Existing data models rely on higher, more complex, and diverse building blocks to represent real life knowledge. For example, in relational databases, we have the concepts: table, row, column, field, primary key, and foreign key. In graph-based knowledge representations, we have nodes and an unlimited number of relationship types. These concepts are "hard-coded" because we cannot modify or advance them, nor reason about their properties within the same framework.
We can highlight 3 major areas for applications of the Datum Universe:
The Datum Universe model is a knowledge representation scheme. It specifies how to represent data and how to make inferences.
Identify the fundamental elements of knowledge and the fundamental elements of intelligence. The goal is to provide a unified minimalistic framework that has the building blocks for representing any kind of knowledge constructs
is the ability to deduce attributes of a datum based on attributes of its "is" relations. For example, if we have "apple is red", and "apple1 is apple", then we can conclude that apple1 is red.
is the opposite of inheritance; deducing properties of a datum using the attributes of its instances. For example, if all of the instances of apple that we have seen are sweet, i.e., "apple1 is sweet", "apple2 is sweet", etc., then we can generalize that all apples are sweet and conclude that "apple is sweet".
is the creation of a new datum to reduce the number of relations. For example, if we have 10 instances of apple, that are all sweet and fresh, and as a result we have many repeating pairs like "apple1 is sweet", "apple1 is fresh", "apple2 is sweet", "apple2 is fresh", etc. A new datum X may be induced as follows "X is sweet", "X is fresh" and "apple1 is X", "apple2 is X", etc. This reduces the number of relations from 2*10 to 2+10.
Knowledge Representation Technology
The Datum Universe represents knowledge using two fundamental concepts:
The Datum; which is an abstract element defined entirely by its relations to other datums.
The "is" relation; which is a directed link between two datums and is the only type of relation allowed among datums.
Intelligence emerges from two fundamental properties of the Datum Universe:
The transitivity property of the "is" relationship (if a is b and b is c, then a is c) allows for inheritance and generalization. See side notes for examples.
Induction; which is the capability of the datum universe to create new datums autonomously. The induced datums reduce the number of connections and provide a built-in classification process. See side notes for an example.
The Datum Universe approach is to build these complex concepts as "soft-coded" constructs out of the minimum "hard-coded" elements. This approach makes it possible to: